ool.languagetool-core.5.7.source-code.ml_server.proto Maven / Gradle / Ivy

Go to download

Show more of this group Show more artifacts with this name
Show all versions of languagetool-core Show documentation

LanguageTool is an Open Source proofreading software for English, French, German, Polish, Romanian, and more than 20 other languages. It finds many errors that a simple spell checker cannot detect like mixing up there/their and it detects some grammar problems.

There is a newer version: 6.5

Show newest version

syntax = "proto3";
package lt_ml_server;
option java_package = "org.languagetool.rules.ml";
option java_outer_classname = "MLServerProto";


service MLServer {
    rpc Match (MatchRequest) returns (MatchResponse) {}
}

// for e.g. resorting suggestions
service PostProcessingServer {
    rpc Process(PostProcessingRequest) returns (MatchResponse) {}
}

message PostProcessingRequest {
    repeated string sentences = 1; // input text to be analyzed
    repeated MatchList matches = 2; // matches of input text, to be transformed
    bool inputLogging = 3; // allow logging of input on error
    repeated int64 textSessionID = 4; // session ID, for partial rollout & A/B tests
}

message MatchRequest {
    repeated string sentences = 1; // input text to be analyzed
    bool inputLogging = 2; // allow logging of input on error
    repeated int64 textSessionID = 3; // session ID, for partial rollout & A/B tests
}

message MatchResponse {
    repeated MatchList sentenceMatches = 1; // results for each corresponding sentence in MatchRequest (aligned)
}

message MatchList {
    repeated Match matches = 1; // list of matches
}

message Match {
    uint32 offset = 1; // start position in the sentence (i.e. not relative to whole text)
    uint32 length = 2; // length of matched area; >0
    string id = 3; // prefix for rule ids; should use a common value for one model/server
    string sub_id = 4; // more specific suffix for rule ids; i.e. one model/server can return multiple values to distinguish between mach types
    repeated string suggestions = 5; // legacy; use suggestedReplacements
    // added later, optional (can be substituted by Java)
    string ruleDescription = 6; // description of the rule; can be specific to rule_id + sub_id or shared for all sub_ids of one rule_id
    string matchDescription = 7; // description of the match; displayed in e.g. the add-on pop-up
    string matchShortDescription = 8; // shortened description of the match; displayed in e.g. LibreOffice context menu
    string url = 9; // to show a link, e.g. with further explanations for this error
    repeated SuggestedReplacement suggestedReplacements = 10; // new and extended suggestions; overwrite any values set in suggestions
    bool autoCorrect = 11; // is this correction eligible for auto-correction?
    enum MatchType {
        /** Spelling errors, typically red. */
        UnknownWord = 0;
        /** Style errors, typically light blue. */
        Hint = 1;
        /** Other errors (including grammar), typically yellow/orange. */
        Other = 2;
    }
    MatchType type = 12;
    sint32 contextForSureMatch = 13;
    Rule rule = 14;
}

message Rule {
    // id, subId, description, url already in Match
    string sourceFile = 1;
    // see http://www.w3.org/International/multilingualweb/lt/drafts/its20/its20.html#lqissue-typevalues
    string issueType = 2;
    bool tempOff = 3;
    RuleCategory category = 4;
    bool isPremium = 5;
    enum Tag {
        picky = 0;
    }
    repeated Tag tags = 6;
}

message RuleCategory {
    string id = 1;
    string name = 2;
}


message SuggestedReplacement {
    string replacement = 1;
    string description = 2;
    string suffix = 3; // Value shown in the UI after the replacement (but not part of it).
    float confidence = 4; // from 0 (lowest) to 1 (highest)
    enum SuggestionType {
        Default = 0;
        Translation = 1;
        Curated = 2;
    }
    SuggestionType type = 5;
}